|
3 | 3 | #include <algorithm> |
4 | 4 | #include <cstring> |
5 | 5 | #include <iostream> |
| 6 | +#include <type_traits> |
| 7 | +#include <unordered_set> |
6 | 8 | #include <vector> |
7 | 9 | #include "Metadata.h" |
8 | 10 | #include "MetadataReader.h" |
|
16 | 18 | constexpr uint64_t kFNV64OffsetBasis = 14695981039346656037ull; |
17 | 19 | constexpr uint64_t kFNV64Prime = 1099511628211ull; |
18 | 20 |
|
19 | | -uint64_t hashBytesFnv1a(const void* data, size_t size) { |
| 21 | +uint64_t hashBytesFnv1a(const void* data, size_t size, |
| 22 | + uint64_t seed = kFNV64OffsetBasis) { |
20 | 23 | const auto* bytes = static_cast<const uint8_t*>(data); |
21 | | - uint64_t hash = kFNV64OffsetBasis; |
| 24 | + uint64_t hash = seed; |
22 | 25 | for (size_t i = 0; i < size; i++) { |
23 | 26 | hash ^= static_cast<uint64_t>(bytes[i]); |
24 | 27 | hash *= kFNV64Prime; |
25 | 28 | } |
26 | 29 | return hash; |
27 | 30 | } |
28 | 31 |
|
| 32 | +MDTypeKind canonicalizeSignatureTypeKind(MDTypeKind kind) { |
| 33 | + switch (kind) { |
| 34 | + case mdTypeAnyObject: |
| 35 | + case mdTypeProtocolObject: |
| 36 | + case mdTypeClassObject: |
| 37 | + case mdTypeInstanceObject: |
| 38 | + case mdTypeNSStringObject: |
| 39 | + case mdTypeNSMutableStringObject: |
| 40 | + return mdTypeAnyObject; |
| 41 | + default: |
| 42 | + return kind; |
| 43 | + } |
| 44 | +} |
| 45 | + |
| 46 | +template <typename T> |
| 47 | +void appendIntegralToHash(uint64_t* hash, T value) { |
| 48 | + using Unsigned = typename std::make_unsigned<T>::type; |
| 49 | + Unsigned unsignedValue = static_cast<Unsigned>(value); |
| 50 | + for (size_t i = 0; i < sizeof(Unsigned); i++) { |
| 51 | + const uint8_t byte = static_cast<uint8_t>((unsignedValue >> (i * 8)) & 0xFF); |
| 52 | + *hash = hashBytesFnv1a(&byte, sizeof(byte), *hash); |
| 53 | + } |
| 54 | +} |
| 55 | + |
| 56 | +bool appendMetadataSignatureHash(MDMetadataReader* reader, |
| 57 | + MDSectionOffset signatureOffset, |
| 58 | + std::unordered_set<MDSectionOffset>* activeSignatures, |
| 59 | + uint64_t* hash); |
| 60 | + |
| 61 | +bool appendMetadataTypeHash(MDMetadataReader* reader, MDSectionOffset* offset, |
| 62 | + std::unordered_set<MDSectionOffset>* activeSignatures, |
| 63 | + uint64_t* hash) { |
| 64 | + if (reader == nullptr || offset == nullptr || hash == nullptr || |
| 65 | + activeSignatures == nullptr) { |
| 66 | + return false; |
| 67 | + } |
| 68 | + |
| 69 | + const MDTypeKind kindWithFlags = reader->getTypeKind(*offset); |
| 70 | + *offset += sizeof(MDTypeKind); |
| 71 | + const MDTypeKind rawKind = |
| 72 | + static_cast<MDTypeKind>((kindWithFlags & ~mdTypeFlagNext) & |
| 73 | + ~mdTypeFlagVariadic); |
| 74 | + |
| 75 | + appendIntegralToHash<uint8_t>(hash, 0xB0); |
| 76 | + const MDTypeKind canonicalKind = canonicalizeSignatureTypeKind(rawKind); |
| 77 | + appendIntegralToHash<uint8_t>(hash, static_cast<uint8_t>(canonicalKind)); |
| 78 | + |
| 79 | + switch (rawKind) { |
| 80 | + case mdTypeArray: |
| 81 | + case mdTypeVector: |
| 82 | + case mdTypeExtVector: |
| 83 | + case mdTypeComplex: { |
| 84 | + const auto arraySize = reader->getArraySize(*offset); |
| 85 | + *offset += sizeof(uint16_t); |
| 86 | + appendIntegralToHash<uint16_t>(hash, arraySize); |
| 87 | + if (!appendMetadataTypeHash(reader, offset, activeSignatures, hash)) { |
| 88 | + return false; |
| 89 | + } |
| 90 | + break; |
| 91 | + } |
| 92 | + |
| 93 | + case mdTypeStruct: { |
| 94 | + const auto structOffset = reader->getOffset(*offset); |
| 95 | + *offset += sizeof(MDSectionOffset); |
| 96 | + appendIntegralToHash<MDSectionOffset>(hash, structOffset); |
| 97 | + break; |
| 98 | + } |
| 99 | + |
| 100 | + case mdTypeClassObject: { |
| 101 | + auto classOffset = reader->getOffset(*offset); |
| 102 | + *offset += sizeof(MDSectionOffset); |
| 103 | + bool hasNext = (classOffset & mdSectionOffsetNext) != 0; |
| 104 | + while (hasNext) { |
| 105 | + auto protocolOffset = reader->getOffset(*offset); |
| 106 | + *offset += sizeof(MDSectionOffset); |
| 107 | + hasNext = (protocolOffset & mdSectionOffsetNext) != 0; |
| 108 | + } |
| 109 | + break; |
| 110 | + } |
| 111 | + |
| 112 | + case mdTypeProtocolObject: { |
| 113 | + bool hasNext = true; |
| 114 | + while (hasNext) { |
| 115 | + auto protocolOffset = reader->getOffset(*offset); |
| 116 | + *offset += sizeof(MDSectionOffset); |
| 117 | + hasNext = (protocolOffset & mdSectionOffsetNext) != 0; |
| 118 | + } |
| 119 | + break; |
| 120 | + } |
| 121 | + |
| 122 | + case mdTypePointer: |
| 123 | + if (!appendMetadataTypeHash(reader, offset, activeSignatures, hash)) { |
| 124 | + return false; |
| 125 | + } |
| 126 | + break; |
| 127 | + |
| 128 | + case mdTypeBlock: |
| 129 | + case mdTypeFunctionPointer: { |
| 130 | + const auto nestedSignatureOffset = reader->getOffset(*offset); |
| 131 | + *offset += sizeof(MDSectionOffset); |
| 132 | + if (nestedSignatureOffset != MD_SECTION_OFFSET_NULL) { |
| 133 | + const auto nestedAbsoluteOffset = |
| 134 | + reader->signaturesOffset + nestedSignatureOffset; |
| 135 | + if (!appendMetadataSignatureHash(reader, nestedAbsoluteOffset, |
| 136 | + activeSignatures, hash)) { |
| 137 | + return false; |
| 138 | + } |
| 139 | + } |
| 140 | + break; |
| 141 | + } |
| 142 | + |
| 143 | + default: |
| 144 | + break; |
| 145 | + } |
| 146 | + |
| 147 | + appendIntegralToHash<uint8_t>(hash, 0xBF); |
| 148 | + return true; |
| 149 | +} |
| 150 | + |
| 151 | +bool appendMetadataSignatureHash(MDMetadataReader* reader, |
| 152 | + MDSectionOffset signatureOffset, |
| 153 | + std::unordered_set<MDSectionOffset>* activeSignatures, |
| 154 | + uint64_t* hash) { |
| 155 | + if (reader == nullptr || hash == nullptr || activeSignatures == nullptr) { |
| 156 | + return false; |
| 157 | + } |
| 158 | + |
| 159 | + if (activeSignatures->find(signatureOffset) != activeSignatures->end()) { |
| 160 | + appendIntegralToHash<uint8_t>(hash, 0xEE); |
| 161 | + return true; |
| 162 | + } |
| 163 | + activeSignatures->insert(signatureOffset); |
| 164 | + |
| 165 | + MDSectionOffset offset = signatureOffset; |
| 166 | + const MDTypeKind returnTypeKind = reader->getTypeKind(offset); |
| 167 | + bool next = (returnTypeKind & mdTypeFlagNext) != 0; |
| 168 | + const bool isVariadic = (returnTypeKind & mdTypeFlagVariadic) != 0; |
| 169 | + |
| 170 | + appendIntegralToHash<uint8_t>(hash, 0xA0); |
| 171 | + appendIntegralToHash<uint8_t>(hash, isVariadic ? 1 : 0); |
| 172 | + |
| 173 | + if (!appendMetadataTypeHash(reader, &offset, activeSignatures, hash)) { |
| 174 | + activeSignatures->erase(signatureOffset); |
| 175 | + return false; |
| 176 | + } |
| 177 | + |
| 178 | + uint32_t argCount = 0; |
| 179 | + while (next) { |
| 180 | + const MDTypeKind argTypeKind = reader->getTypeKind(offset); |
| 181 | + next = (argTypeKind & mdTypeFlagNext) != 0; |
| 182 | + if (!appendMetadataTypeHash(reader, &offset, activeSignatures, hash)) { |
| 183 | + activeSignatures->erase(signatureOffset); |
| 184 | + return false; |
| 185 | + } |
| 186 | + argCount++; |
| 187 | + } |
| 188 | + |
| 189 | + appendIntegralToHash<uint32_t>(hash, argCount); |
| 190 | + appendIntegralToHash<uint8_t>(hash, 0xAF); |
| 191 | + |
| 192 | + activeSignatures->erase(signatureOffset); |
| 193 | + return true; |
| 194 | +} |
| 195 | + |
29 | 196 | } // namespace |
30 | 197 |
|
31 | 198 | // Essentially, we cache libffi structures per unique method signature, |
@@ -267,11 +434,13 @@ uint64_t hashBytesFnv1a(const void* data, size_t size) { |
267 | 434 | rvalue = malloc(cif.rtype->size); |
268 | 435 | rvalueLength = cif.rtype->size; |
269 | 436 |
|
270 | | - const size_t signatureLength = static_cast<size_t>(offset - signatureStart); |
271 | | - if (signatureLength > 0) { |
272 | | - const auto* signatureBytes = |
273 | | - reinterpret_cast<const uint8_t*>(reader->data) + signatureStart; |
274 | | - signatureHash = hashBytesFnv1a(signatureBytes, signatureLength); |
| 437 | + if (signatureStart != MD_SECTION_OFFSET_NULL) { |
| 438 | + uint64_t canonicalSignatureHash = kFNV64OffsetBasis; |
| 439 | + std::unordered_set<MDSectionOffset> activeSignatures; |
| 440 | + if (appendMetadataSignatureHash(reader, signatureStart, &activeSignatures, |
| 441 | + &canonicalSignatureHash)) { |
| 442 | + signatureHash = canonicalSignatureHash; |
| 443 | + } |
275 | 444 | } |
276 | 445 | } |
277 | 446 |
|
|
0 commit comments